Forensic Marketing Is Real: A Case Study in Superhero Analytics
It should be noted that the name of the customer has been changed to protect my income. All marketing tools are assumed ineffective until proven otherwise in a court of the laws of mathematics.
Okay, that was a bit of a nerdy statement, but necessary. I do protect my clients’ privacy when consultations are needed, and thus I protect my income. I also take the approach that something in a clients’ marketing machine is wrong, that somewhere in their marketing plan there is a butler holding a candlestick in the library … with Col. Mustard.
On one particular afternoon, I got the call that a large global technology company (whom we shall refer to as Big Tech) had an issue. This particular company had taken a loss in sales in one of their Asian markets. The year-over-year sales on the small and medium business website had declined by 40 percent. The top-level executives wanted answers and possibly a recommendation on how to fix the issue. At times, this job can make one feel a little like a detective, or a superhero. In this instance, I didn’t slide down a pole hidden behind a bookcase into my secret lair and climb into my Statmobile. Nope, much more exciting—I fired up my laptop.
This was within a year or two of the 2008 downturn when a lot of businesses’ sales had slowed. Big Tech had seen a slow in sales, much the same as everyone else. One company cannot change global economic conditions; that’s out of their control. What the client wanted wasn’t necessarily a super fix, but rather to establish what was within their control and what was not. The waters can sometimes get a bit muddy without supportable facts and that means numbers.
After a few interviews with Big Tech’s executives and sales people, it became apparent that the loss was being blamed on the U.S. dollar’s poor performance at that time as well as the economy. The thought process in place that supported this theory was that if the dollar was trading badly against the yen, U.S. companies would be doing poorly abroad. I was not opposed to the idea, but I was not convinced either. With most large clients, such as Big Tech, you often get little guidance as to where to look and how to approach a question. For analysts like myself, this is a godsend. We will generally have an open, green field to begin our study. So there I was with all of Big Tech’s raw Internet data and carte blanche to do what I had to do to get some answers.
The first hurdle one must overcome when approaching a big chunk of data like the one that Big Tech gave me was deciding where to begin. Is it better to start with acquisitions? Do you look at the site and how customers are interacting with it, or do you get right to where the meat meets the metal and dive into the cart check-out data? Time to do some data mining and correlating.
Using data mining techniques that are now built into Adobe Analytics Premium, I was able to find a strong correlation between email recipients who landed on a general product page and those who landed on a specific product page. For instance, if the customer clicked on a widget that was advertised in an email blast and landed on the product page, they most likely would purchase said product. Those who landed on a general page and were forced to look for the advertised product more or less just gave up. At least that is the inference that could be drawn from the data. Using predictive modeling techniques, we were able to test the theory offline and it looked like that dog would hunt.
By simply changing the URLs in the email blasts so that all the redirects pointed toward the products being advertised, we were able to see an immediate impact. This simple “fix” landed Big Tech an overnight revenue increase of $150,000 to as much as $500,000 in weekly incremental revenue. This was a huge win for Big Tech because they finally had gotten a handle on what they could control instead of feeling victimized by the things that they could not. The job was still not complete, however.
I went back and decided to look at the exchange rate theory to see if there was any credence to the idea that a poorly performing dollar was a factor. In analytics, one never discounts a theory; I’ve seen some pretty crazy correlations that turned out to be spot on (remember knitting and poison?). I still approach each theory with a grain of salt, though. I brought the exchange rate between the dollar and the yen into the equation (literally) and found next to zero correlation between sales and exchange rate. Another win. In addition to email recommendations, senior management was able to disprove a prevalent theory and gain some accountability within their organization.
Did Big Tech close the gap on their 40 percent decline? No, not overnight. At the time, competing products were hitting the market, so its sales would have been impacted anyway. The economy was not in as favorable a condition as it had been either. Both of these issues were contributing factors that were out of Big Tech’s direct control. But the company was able to identify what it could control and maximize its efforts and resources accordingly. I could dust off my britches and call it a day on this case.
The thing to remember is that you do not have to be a Big Data superhero if you have the right tools. Tools such as Adobe Analytics Premium allow you to do the dead reckoning you need to do to get your marketing back on course. Spending more time answering questions and not being overwhelmed with manipulating Big Data is what analytics is about. That, and getting to travel to exotic places like Tokyo … on a conference call.